ESTRO 2021 Abstract Book

S1573

ESTRO 2021

Conclusion Inhomogeneous continuous adaptive dose escalation can significantly increase the dose to gross tumour volume without increasing organs at risk doses or compromising dose to microscopic disease. This may increase local control and overall survival without increasing toxicities or affecting patient’s quality of life.

PO-1845 Regression models for predicting plan parameters of two kinds of hybrid plans for stage III NSCLC H. Wang 1 1 Shanghai Chest Hospital, Shanghai Jiao Tong University, Radiotherapy Department, Shanghai, China

Purpose or Objective To establish regression models of physical and EQD 2

plan parameters of two kinds of hybrid plans for stage III NSCLC.

Materials and Methods Two kinds of hybrid plans named C&S and C&SIB were retrospectively made for 20 patients with stage III NSCLC. Prescription dose of C&S plans was 30×2Gy for PTV LN (lymph node) and sequential 4×12.5Gy for PTV PT (primary tumor), while prescription dose of C&SIB plans was 26×2Gy for PTV LN and sequential 4×2Gy for PTV LN combined with 4×12.5Gy for PTV PT . Regression models of physical and EQD 2 plan parameters were established based on anatomical geometry features for two kinds of hybrid plans. The features were mainly characterized by volume ratio, min distance and overlapping slices thickness of two structures. The possibilities of regression models of EQD 2 plan parameters were verified by spearman’s correlation coefficients between physical and EQD 2 plan parameters, and it was investigated that how the consistence of coefficients of determination ( R 2 ) of regression models between physical and EQD 2 plan parameters was affected by the correlations between physical and EQD 2 plan parameters. Non-transit electronic portal imaging device (EPID) dosimetry analysis was performed to rule out dosimetric errors during hybrid plans delivery. Results Physical and EQD 2 PTV LN CI 60Gy showed a high linear correlation with PTV PT volume, PTV LN volume and min distance(PT to LN). PTV (PT+LN) CI 60Gy showed a linear correlation with PTV PT volume, PTV LN volume, min distance (PT to LN) and overlapping slices thickness (PT and LN) . Total lung and ipsilateral lung V 20 and MLD showed a cubic function with volume ratio (LN to total lung) and volume ratio (LN to ipsilateral lung) . Heart D 5 , D 30 and MHD would be more susceptible to overlapping structure (heart and LN) . Min distance (PT to ESO) may be an important feature for predicting esophageal max dose for hybrid plans. It’s feasible for regression models of EQD 2 plan parameters, and the consistence of the fitting goodness of regression models between physical and EQD 2 plan parameters had a positive correlation with spearman’s correlation coefficients between physical and EQD 2 plan parameters. The delivery errors of hybrid plans components were acceptable using the criterion of 2%/2mm and passing rate more than 90%. Conclusion The fitting goodness of regression models could be at least moderately perfect in this work, and the models have a potential to predict physical and EQD 2 plan parameters for two kinds of hybrid planning. PO-1846 Experiments to validate a Mid-Position workflow: Results with two independent implementations E. Jackson 1 , D. Boukerroui 2 , M.J. Gooding 2 , P. Remeijer 1 1 Netherlands Cancer Institute, Radiotherapy, Amsterdam, The Netherlands; 2 Mirada Medical Ltd., Science, Oxford, United Kingdom Purpose or Objective The Mid-Position method accounts for breathing during RT by using the 4DCT scan to create a time-averaged 3DCT image and patient specific margins that can be used for planning. Margins are calculated from estimates of tumor motion to define a PTV that covers the tumor throughout the breathing cycle. This results in smaller margins than those using the ITV method[1]. The purpose of this work is to describe a set of experiments that can be utilised to validate an implementation of the Mid- P algorithm by comparing Mid-P outputs against independent measures of breathing motion and HU intensity within the GTV. Materials and Methods Mid-P images and estimates of peak-to-peak motion amplitudes within the GTV were calculated from 4DCT images (10 phases, 126 slices at 3mm, 512x512 pixels at 1mm) for 15 patients. The GTVs were manually contoured on Mid-P scans, and PTV expansion margins were computed based on motion estimates according to van Herk et al.[2]. These were compared against motion estimates (i) computed from intensity-weighted centroids of target contours drawn on each phase and (ii) inferred from local rigid registration of ROIs around targets. For 5 patients where Breath-Hold (BH) scans were available, voxel distributions between GTVs in BH images and Mid-P scans were compared. To assess motion outside the GTV, peak-to-peak amplitudes of 41 lung landmarks from the POPI model [3] were compared to Mid-P motion estimates at the same points. Two implementations were compared: Method 1, in use clinically at the NKI[1]; Method 2, an automated research prototype